DanielLeens commented on issue #13321:
URL: https://github.com/apache/seatunnel/issues/13321#issuecomment-6983643297

   Confirmed the corrupt-candidate abort at the method level on 
`89f91a89af15f33d97a942d8b65f4fc1b0c80bf5` (macOS ARM64, Java 11.0.27).
   
   Reproduction steps:
   1. Serialize a valid `CompletedCheckpoint` (ID 1, ordinary checkpoint type) 
with the **real `ProtoStuffSerializer`**, and return its `PipelineState` from 
the storage dependency.
   2. Invoke actual `CheckpointManager.getLatestCheckpointStateByType("7", "1", 
RestoreMode.CHECKPOINT)`. The healthy control selects ID 1.
   3. Return the same healthy candidate together with newer ID 2 whose state is 
truncated bytes `{10, (byte) 255}`.
   4. Invoke the same production method. It throws `RuntimeException` with the 
`Truncated` deserialization failure instead of selecting ID 1.
   
   [Executable Java probe and exact build/run 
commands](https://github.com/apache/seatunnel/issues/13313#issuecomment-6983599115):
 see `issue13321()` in `CheckpointIssuesRepro.java`.
   
   This is an in-process storage fault-injection test with unmodified 
production code, not an actual disk-corruption or cluster-failover experiment. 
It verifies the current lack of per-candidate fallback; it does not establish 
the safety/semantics of any proposed fallback policy.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to